Commit messages

August 20, 2014

mhunter committed r3768 in OpenMx at 15:00:
Modify .svnignore files.  WLS branch and trunk changes.  Added continuous only wls test.  Found and added state space algebra test.
in /:

July 31, 2014

jpritikin committed r3705 in OpenMx at 11:47:
Implement new unprotect strategy

The old idiom was

// do stuff
Rf_unprotect(1);  //foo

The second line expresses the hope that we are unprotecting the same
thing that we protected. However, this can fail when "do stuff" involves
anything non-trivial. If there is a protect (or worse, an extra
unprotect) inside "do stuff" then the unprotect will not unprotect foo
but something else.

The new strategy is to check whether the protect stack is
at the same place where we protected, ensuring that our
protects are always balanced by unprotects. It is still
required to think about what is necessary to protect,
but now there is less chance of something going wrong
in the unprotect part.
in /trunk:

July 7, 2014

jpritikin committed r3616 in OpenMx at 03:34:
Add a conformability checking step to the backend 1/2
in /trunk:

May 10, 2014

tbates committed r3445 in OpenMx at 11:27:
minor error msg mod to suggest action if non-square matrix declared as cov: perhaps you meant type = 'raw' instead of 'cov'?
in /:

May 6, 2014

mhunter committed r3425 in OpenMx at 17:52:
Confidence Intervals now give the name of the parameter if they are from an MxMatrix instead of the MatrixName[row,col].  Confidence intervals on algebras still give AlgebraName[row,col].
in /:

April 28, 2014

jpritikin committed r3379 in OpenMx at 08:08:
Delete unused variable
in /trunk/src: omxRFitFunction.cpp (modified, previous: r3360)

April 17, 2014

jpritikin committed r3360 in OpenMx at 11:43:
Remove unused arguments from omxRaiseError, omxRaiseErrorf

March 12, 2014

jpritikin committed r3138 in OpenMx at 10:46:
Enable R_NO_REMAP for a cleaner namespace
in /trunk:

December 31, 2013

mhunter committed r3035 in OpenMx at 17:38:
Update copyright.
in /trunk:

August 17, 2013

jpritikin committed r2851 in OpenMx at 00:13:
Allow ComputeIterate to test maximum absolute change

+ The adjustStart flag is gone. There is no longer a reason to request
adjustStart=FALSE. It might do something else besides adjust starting
values so the name isn't accurate either.

+ mxComputeOnce no longer has a fit=TRUE default action. By default it
does nothing.

+ Not evaluating the fit helps estimate IFA models. The performance is
about the same. The main advantage is that some bulky caching can be
turned off with no penalty. Here is page fault data for ifa-cai2009.R.
Ignore the difference in runtime because the EM cycle tolerance was not
equated between runs.

With cache:

55.60user 0.07system 0:08.41elapsed 661%CPU (0avgtext+0avgdata 83788maxresident)k
0inputs+0outputs (0major+72249minor)pagefaults 0swaps

Without cache:

60.64user 0.10system 0:09.01elapsed 673%CPU (0avgtext+0avgdata 71276maxresident)k
0inputs+0outputs (0major+61669minor)pagefaults 0swaps
in /trunk:
Syndicate content